NEW (pseudo-) random number generator in 5.18?
I hope mathematically-knowlegable programmers give that a VERY careful review, as it underpins cryptographic programs. Just sayin' ...
Linus Torvalds has released version 5.17 of the Linux kernel. "So we had an extra week of at the end of this release cycle, and I'm happy to report that it was very calm indeed," Torvalds wrote in his weekly state of the kernel post. "We could probably have skipped it with not a lot of downside, but we did get a few last- …
>>Version 5.18 of the kernel is tipped to feature the debut of Intel's plans for "software-defined silicon" – a tech about which Intel has remained virtually silent, other than hints on mailing lists about features that would allow payments to enable different features in processors.
How do you feature something we know nothing about, well your guess is as good as mine?
I'm confused about the workings of the business model.
I start by assuming all the dies coming off a wafer are identical and the method of enabling is universal (if not how are these things made?) So if I buy microcode it must surely work on all the same devices. How do Intel prevent unauthorized distribution, intentional or otherwise?
If there is a method of prevention it must be something like TPM on a stick.
I don't think this can be anything to do with GPL versions as the source code is readable in both. This has got to be about binary blobs.
Yes something like TPM on a stick is it but remember, even if it turns out that someone hacks it and can turn on the paid-for feature for diddly squat, it won't hurt Intell too badly because they hope to milk the large customers who can't afford to be seen to deploy H4xXoR tricks.
Business model Revenue Stream xAAS BS bingo aside: I hate that sort of thing it seems spiteful to actually give some hardware to a customer and say: don't use it it's not allowed, it's MINE!!!!.
It's a bit like Adam and Eve and that apple thing: god's mindset seems inexcusable, it's just a trap. Putting any trust in an entity like that is obviously misguided.
It's complicated, but it's really not unethical (at least in my view). It becomes sensible when the cost is predominantly in the development, not the manufacturing. Say there are two features A and B. There are X people who will pay x for both features, and an additional Y people who will pay y (y < x) for just feature A, but who won't pay x. It is quite often the case that Xx > (X+Y)y. If you _don't_ give the Y people feature B, but disable it somehow, you end up with just charging the higher price for the full featured option and the Y people get nothing.
Of course, it is _always_ the case the Xx + Yy >= (X+Y)y, (and equality only if X=0) so this is also a way for greedy manufacturers to extract every last penny - but people _do_ get the benefit of what they pay for.